=== bunny is now known as Guest32615 [15:04] can we have modules that we can drag and drop to easy start off creating apps? [15:04] ubuntu-guy: there's going to be a session about that later today [15:04] ubuntu-guy: http://summit.ubuntu.com/uds-1403/meeting/22136/ubuntu-component-store/ [15:04] is summit down? [15:05] if you're interested in helping us work out the details of that, please join [15:05] kenvandine: not that I know of [15:05] kenvandine: works here [15:05] not loading for me :/ [15:05] thanks will do [15:05] kenvandine: no pages load? [15:05] none [15:05] :/ [15:05] kenvandine: other ubuntu pages load for you? [15:06] oh... there now it loaded [15:06] 10 minutes i couldnt [15:06] did you forget to use your manners? [15:06] it's very picky about that [15:06] where's the hangout link? [15:06] mhall119, i guess :) [15:06] http://summit.ubuntu.com/uds-1403/meeting/22164/appdev-1403-releasing/ [15:07] if youwant to join the hangout, and have something to add, you can join at https://plus.google.com/hangouts/_/hoaevent/AP36tYeyOMZOQu5nuv4KbFTaNK5B5BhUlzky8d1Ql-Yu8jQVjCY38w?authuser=1&hl=en [15:07] please mute yourself when you're not speaking though [15:07] I would say release them when they are ready [15:07] Thanks! === Totalschaedn1997 is now known as Totalschaden1997 [15:11] did you try using qmltestrunner for ui testing of the sdk, when autopilot did not work? [15:11] bzoltan, QUESTION: can you speak a little more about the idea of separating the tools (QtC and etc) from the UI Toolkit stuff? How was it organized? how will it be organized? [15:12] kyleN: do you want to join the hangout? [15:13] ChrisGagnon: not yet, but soon we will investigate other options [15:13] did the video start yet? I don't see it [15:14] t1mp: it has [15:15] Kaleo, yes that made sense [15:15] yes [15:15] can you mention frameworks specifically with APIs [15:15] kyleN, can you elaborate? [15:16] I still don't see video on http://summit.ubuntu.com/uds-1403/meeting/22165/appdev-1403-uitk/ anyone else? [15:16] argh [15:16] the concepts of frameworks remains somewhat vague. yet devs need to state it in their manifest. how does stating it in the manifest relate to API selection at runtime and etc [15:16] wrong session page. that's tomorrow :) [15:18] good, that is the explanation I was looking for [15:19] current framework is there with some exceptions, apis that need services or hardware thats no available [15:19] that would work only if there's some api validator to see if nothing outside of the framework is used [15:19] you should join the hangout ;) [15:21] sergiusens: that would be done in the SDK, it would have to have separate API definitions that it uses to validate code [15:21] mhall119, so if I build my c++ plugin and that uses stuff outside of the framework; would that be checked against as well? [15:22] QUESTION: can you discuss the separation of DEV ppa from RELEASED ppa and perhaps mention testing and other steps that may be taken to ensure RELEASED PPA quality [15:24] and will code completion work for HTML5/JS apps too? [15:25] jono_, bzoltan that depends on a Qt5 for x86 built with gles [15:25] sergiusens: stuff outside the framework should be in your project root, so yes it should [15:26] hack arounds :-) [15:27] pad: http://pad.ubuntu.com/uds-1403-appdev-1403-releasing [15:28] jono_, that's being worked on; we need to multibuild qt5 [15:28] too much code has armhf == gles and !armhf = gl [15:29] ew [15:29] sergiusens: and that's in upstream qt? [15:30] QUESTION: will code completion in QtC work with HTML5/JS apps? [15:30] heh, my phone has working alarms now....evidently I'll be getting a notice before all my calendar events :) [15:32] mhall119, packaging and almost anything tht compiles and links against qt5 that we use [15:33] mhall119, these: https://launchpad.net/~rsalveti/+archive/qt-gles-test [15:33] jono_: https://blueprints.launchpad.net/ubuntu/+spec/appdev-1403-releasing [15:37] kyleN: there will be a session 16:00 - 16:55 UTC HTML5 SDK Roadmap ( App Development ) I would ask that from alex-abreu [15:38] bzoltan, ok. it seems to span both domains: SDK and HTML5 runtime [15:39] is the X86 emulator a DESKTOP or a phone or a tablet? [15:39] i think so [15:39] oops, server error... [15:40] is raring still supported at all?! [15:40] weird I got kicked out [15:40] and cant rejoin [15:40] lool, raring shouldn't be supported [15:40] raring is not supported anymore; quantal is still up for a couple of months [15:41] jono_: try rejoining using your own account instead of UbuntuOnAir? [15:41] you can't even dput to raring anymore [15:41] did I get kicked out? [15:41] bzoltan: well let's not screw them but let's not spend time supporting them, the support path for them is to upgrade IMO [15:41] Kaleo, seems you have [15:41] anyway, completely off topic, sorry :-) [15:41] lool, but raring is EOL, you can't dput to it [15:42] sergiusens: well exactly [15:42] mhall119, tried, I can't get in [15:42] lool, so it's impossible to support for real :-) [15:42] jono_: huh, I can't even eject your frozen connection [15:42] odd [15:42] I'm surprized launchpad still accepts raring uploads to PPAs [15:42] jono_: can you watch the video at least? [15:43] mhall119, the stream only shows pmcgowan bzoltan and yourself [15:43] lool, it doesn't; well I couldn't [15:44] QUESTION: now that we separate Qtc and tools fomr API, can we stop using "SDK" to include runtime stuff? [15:44] SDK should be QtC and tools, I think [15:44] D=Development, not runtime [15:44] SDK includes the emulator images in my view [15:44] which are test targets [15:45] but the sdk runtime is in the images only as you say [15:45] mhall119, its ok, I think I got most of what I want out of the session [15:45] ok, food for thought anyway [15:45] sounds like bzoltan has got it well handled [16:00] dbarth, do you need help setting up the event? [16:00] dholbach, whats the hg ? [16:01] alex-abreu, mhall119 said he had asked dbarth if he could run it - so I asked dbarth now :) [16:01] alex-abreu, I'm happy to set it up [16:02] ok, doing it now - just a sec [16:02] https://plus.google.com/hangouts/_/hoaevent/AP36tYfdCLyIpbF9imaNkUmbmkfITWHBxtfOqmxvAPxHhZlCIY8NIw [16:03] ok, summit is updated [16:03] alex-abreu, dbarth, kyleN: ^ [16:04] and we are live [16:04] dholbach: ok [16:04] dholbach: just arrriving, hang on [16:04] dholbach: ant, kylen i guess [16:05] I'm here and listening guys [16:05] antdillon, cool [16:05] antdillon, great [16:05] antdillon, hopin if you want [16:06] does anyone else want to join in ? [16:06] blueprint is available here: https://blueprints.launchpad.net/ubuntu/+spec/appdev-1403-html5-sdk [16:06] notes are going to be taken here: http://pad.ubuntu.com/uds-1403-appdev-1403-html5-sdk [16:08] dholbach: https://docs.google.com/a/canonical.com/presentation/d/1cxdhwBJwaTikDQ_qFFCSLDllFG1u6_LOkq9PiLO5INg/edit#slide=id.g1d034350e_00 [16:14] any questions? any feedback? anyone who wants to join the session? [16:18] is anyone taking notes? [16:18] QUESTON: what frameworks are you considering for responsive layouts? Pure? Bootstrap? etc... [16:18] there's no any question. [16:19] kyleN, responsive isn't really about portrait vs landscape; it's about screen sizes. A screen isn't landscape; it's just X px wide vs x/3 px tall :) [16:19] media queries work fine -- don't invent a new responsive thing. Anyone developing HTML stuff should already know about this stuff ;) [16:19] i'm taking notes here: http://pad.ubuntu.com/uds-1403-appdev-1403-html5-sdk [16:20] QUESTION: with cordova framework can we do wrappers to detect any device pixel format? [16:20] more spcific [16:20] specific [16:21] when we have a touch or laptop can we detect it [16:22] kyleN, the sidestage doesn't matter. Your app should say "hey, I am in a window which is of size X by Y" [16:23] ya, respond to the amount of real estate that you have [16:23] totally right :) [16:24] Making a bootstrap theme, and making sure that bootstrap works really well, would be ideal [16:25] i am here [16:25] o/ [16:25] hi. okay [16:26] astronfestmon: just coming to your question next [16:26] okay. thanks [16:26] aquarius: up for it? [16:26] daker: hey [16:26] heh, daker and I were discussing translations the other day; the Moz b2g stuff, and jsgettext [16:27] dbarth, up for what? [16:27] bootstrap [16:27] ? [16:27] dbarth, I don't want to promise tht I'll have time for anything :( [16:28] QUESTION: can we review the work items from last time (https://blueprints.launchpad.net/ubuntu/+spec/appdev-1311-html5-sdk) and see if there's still work to be done or stuff to be landed? [16:28] QUESTION: feedback from another session: "Update Cordova template to be more Ubuntu-like (e.g replace the Cordova icon, make it do something useful, can be very simple though)" - can we make this a work item too or should it be a bug report? [16:29] can you all help taking notes? http://pad.ubuntu.com/uds-1403-appdev-1403-html5-sdk - thanks! [16:29] aquarius: nw [16:29] I don't understand why compatibility with the rest of the Ubuntu project is useful, for gettext. App developers are not likely to use Launchpad and rosetta [16:30] it's api [16:30] If we are talking about DPI we can use media queries to get that [16:30] astronfestmon, are you talking about DPR? [16:30] yes i am [16:30] aquarius, on the other hand, using Launchpad for translations is VERY nice and is where translators (loco teams) look at [16:31] astronfestmon, not atm ... but this is something that we could include, there will be a runtime js binding [16:31] dholbach, yeah, but we are not exactly encouraging app developers to use launchpad these days [16:31] well we could do that in our docs [16:31] I dont think m,y mic is working [16:31] I was just bringing it up as a side-point :) [16:31] antdillon, we need to be very sure that all devices accurately report their details to the platform so that the platform can correctly do media queries, mind [16:32] if we use media queries it's assynchronous right [16:33] astronfestmon: yes, like pretty much everything in the JS engine i guess [16:33] okay. thanks [16:33] astronfestmon: you have concerns about that? [16:33] yes a little [16:34] because the media queries are not my strength inside an app [16:34] aquarius, Sure, I agree it all system settings accessible via the api would be great [16:34] aquarius, But for layout and styling media queries would work [16:35] antdillon: but should we use dpi/media queries, or rather rely on the grid unit system preferably? [16:35] On code completion, etc, building in jshint and htmllint would be great, because that will pick up on lots of thnigs [16:35] ie, dpi is low-level and maybe important for fullscreen apps / games [16:36] but for regular UIs, shouldn't GU be prefered? [16:36] the issue with DPI's is that you need it to know what resolution of static images to use [16:37] that's why you need some representation of dpi [16:37] QUESTION: if we use dpi for games WEBGL is better than a UI stylish for it. Or Canvas? [16:37] dbarth, A use case for dpi media queries is for responsive images [16:37] man, seriously, half of what you're describing on the hangout is handled by jshint :) [16:39] antdillon: ok [16:39] aquarius: jshint, nice [16:40] fine, it doesn't offer type tracking, but that's because YOU DON't NEED TYPE TRACKING because this isn't C ;-) [16:40] It would be good to run this on submit at least [16:41] aquarius, we need to validate the API calls & stuff like that [16:41] aquarius, and have completion [16:41] that's the gist of any completion tool [16:41] alex-abreu, no, no we don't. Nobody else does ;) [16:41] either dynamic eval & stuff [16:41] or somesort of type hiints [16:41] code completion is fine -- that's what TAGS files are for and so on, right? [16:41] aquarius, wrong :) ... check out e.g. codemirror [16:41] * aquarius laughs [16:42] I like Sublime Text, myself :) [16:42] aquarius, I am not sure that etags is something that I'd like to see again :) [16:42] no tags ! :) [16:42] Are we not going to use grid units for HTML5? [16:42] I honestly don't think that we need intellisense [16:42] I'm with aquarius :) [16:42] aquarius, thats the thing ... no intellisense, but at least something interesting [16:42] I can't see how we can use grid units for HTML5, becuase the CSS parser doesn't know about them [16:42] I think the grisd is important [16:42] aquarius, as much as we can get :) [16:42] grid* [16:43] unless we *insist* that all the CSS is actually SASS [16:43] Do we have an example app packaged? [16:43] yes i think that about SASS [16:43] alex-abreu, I see your point about not using etags, etc, yeah; I just don't want someone to spend a million years trying to write a type inferencer for JavaScript :P [16:44] if you need/want a type inference system for js there are tools for that already - TypeScript for example [16:44] I agree with not using grid units (as this would need "inventing") but % grid are common to html developers [16:45] We can use a simple % grid [16:45] in qml can we develop grids with no html tags right? [16:45] antdillon, yeah, certainly, using a grid is a common idea. Bootstrap, 960 grid, etc [16:45] but that doesn't need to be part of the SDK [16:45] If I want to use bootstrap, I can. [16:45] right [16:45] ok [16:45] I would like to use the same naming as bootstrap to help developers [16:45] If we want to use grid units in CSS, we need to wait for var() support. :) https://developer.mozilla.org/en-US/docs/Web/CSS/Using_CSS_variables [16:46] I think the grid should be part of the SDK as we want to bring a common grid layout to apps [16:47] aquarius, yes this is something that is not standardized I think [16:47] antdillon, if the grid is named for bootstrap then I think it should *be* bootstrap. [16:47] At least as a best use case [16:47] I think that so. [16:47] You don't want to lock people into any grid system, grids will not work for all applications [16:47] people need to be able to customize them [16:47] wow. Totally disagree with all of you. If you want to insist that people use Ubuntu-specific technologies in an HTML app, then it's not HTML any more. It's just an Ubuntu-specific layouit language which looks a lot like HTML. [16:47] agreed with hatch [16:48] aquarius, Yes happy to use that I just dont want the devleopers to have to learn a new naming convention for the grid [16:48] aquarius: the point here is to not have html5 developers behind, when it comes to having a native Ubuntu look [16:48] i for one want to develop an app with the nice layout, without (hopefully) to dive into a load of css for that [16:49] but then, we don't want to make that a pre-requisite for runninng your app in the container [16:49] aquarius, I dont mean force them to use it but have it in the SDK to use if the developer doesnt mind which grid they use [16:49] Sadly, you do have to support plenty of web browsers. Because if anyone wants to write an Ubuntu ONLY app, they're going to use QML for it. If they're usnig HTML5 to write an app for Ubuntu, they'll likely be using that app on multiple platforms, [16:50] Good question aquarius, what browsers do we support? [16:50] antdillon, yeah. It's not about browsers you suport -- you support the Ubuntu browser, that's it. [16:51] It's that app developers who are using the Ubuntu engine to deliver apps will likely also be delivering that app on other platforms, so they will not want to add Ubuntu-specific things like grid units to it, because those won't work elsewhere. [16:51] aquarius, I imagined as long as it worked on the latest release of blink its good to use [16:51] (Ubuntu stuff which is *added*, like the Apple meta tags, are fine) [16:52] dbarth, Have you guys worthed on the video player widget? [16:53] The reason for wanting the Media service and the Download service is not so you feel like an Ubuntu app; it's so you can do things in the background :) [16:53] dbarth, I say that because the web team here have just developed a played skinned to match the phones [16:53] why does URL dispatcher need to be accessible to HTML5? Why can't I just do "location.href='someurl://whatever'"? :) [16:54] aquarius that's kind of the hulk smash approach, if it's within your own app you want to use pushstate [16:54] hatch, the url dispatcher is for talking between apps [16:54] Everyone loves a todo app! [16:55] hatch, agreed on pushstate for my UI urls :) [16:55] ohh ok, sorry carryon :) [16:55] Angry birds ... done [16:55] :) [16:55] the most important thing that you guys could do is make Oxide happen. :) [16:56] then things like XHR work properly :P [16:56] You can't build a U1 app without Content Hub and the Download API [16:56] Trust me on this. [16:56] Speaking as the world expert on building U1 apps ;) [16:57] Example app? [16:57] aquarius, we have the contenthub & we will have the download manager [16:57] Once those things exist, it's reasonably easy -- I know this because I've done it about eight times :P [16:57] alex-abreu, yep! [16:57] alex-abreu, do you have the *new* content hub stuff? [16:57] aquarius, not yet, in the works ... :) [16:57] we have the old [16:57] I could probably be talked into writing a U1 HTML5 app. [16:57] * dholbach notes down a work item for aquarius [16:57] since I have done it before about three times :P [16:59] dholbach, go for it [16:59] Im happy with that [16:59] Great thanks guys! [16:59] Ill get my mic working! [16:59] me too [16:59] great chat everyone, thanks! [16:59] thx guys ! [16:59] thank you, chaps [17:00] great input [17:00] inputs [17:00] thank you all [17:00] QUESTION: How do people get more involved in this? [17:00] bye [17:00] aquarius: just noticing your message about u1; can we talkabout that next? [17:00] hatch, hang out with daker and alex-abreu and look for stuff to do :) [17:00] hatch, you can reach us in #ubuntu-webapps [17:00] dbarth, certainly -- happy to talk about it [17:00] see how we can share work here [17:01] hatch, talk to me or yeah daker [17:01] hatch: bzr branch [17:01] hatch: standard ubuntu way [17:01] hatch, we would love to have extra help / inputs [17:01] hatch: you can email us on the webapps list as well, and see on irc #ubuntu-webapps as well [17:01] dbarth, what would you like to know? [17:02] aquarius: review the scope of what it should do [17:02] aquarius: knowing where the dragons live [17:02] that kind of things [17:02] thanks joined the channel [17:02] hatch: see you there [17:02] I'm on the Juju UI team btw [17:03] hatch: ah nice [17:03] this UI is brilliant [17:03] hatch, oh tremendous! [17:03] :-) [17:04] dbarth, log in to U1 to get an oauth token; cache the token; hit the API to retrieve your list of files and folders; cache that data; provide a pagestack to navigate down into folders; refresh button to refresh the cache; (stage 1) allow downloading a file from U1; when you elect to download a file, calculate a signed URL for it and hand that to the contenthub and download daemon. [17:04] dbarth, stage 2: allow sharing files via public url; allow uploads to U1 [17:04] aquarius: sounds like a plan ;) [17:04] but stage 1 would get a lot of people up and running :P [17:04] the first part i remember from previous use of u1db [17:04] dbarth, I wrote a QML U1 app in a day a couple of weeks ago, but I put it on hold until the download and content APIs are available ;) [17:05] so I can do the same with HTML. [17:05] oh i see [17:05] you need those, because it is useless to have the U1 app download files for itself. [17:05] you only ever want to download files in order to give them to some other app :) [17:06] I spent a bunch of time talking to kenvandine about this use case for the content hub api :) [17:06] aquarius, the download manager is def something that I want to hev quickly now that it has reached a stable/usabel state, the rest you have already pretty much [17:06] it needs the New Content Hub Stuff, but that's hopefully coming soon [17:06] alex-abreu, yeah [17:16] alex-abreu, in my content-hub roadmap session, i assigned a work item for you for the bindings update :) [17:16] kenvandine, oh you did :) [17:16] * aquarius laughs [17:16] drive-by work item assignment :P [17:16] kenvandine, when was the session ? during the html5 one ? [17:17] :) [17:17] yes [17:17] aquarius, quiet or i'll assign one to you :) [17:17] great minds [17:17] kenvandine, hey, I was not concentrating and now I have to write a U1 app in HTML5. Which needs the content hub ;) [17:17] alex-abreu, your task is the biggest blocker for our landing... no pressure [17:18] aquarius, woot! [17:18] aquarius: ok, so if you've got most of the u1 app, the simplest way to share is to identify what you miss and we can provide [17:18] aquarius: like content hub [17:18] aquarius: download manager i guess [17:18] aquarius: and you say you don't need grid units [17:18] dbarth, i've already added features to the hub for aquarius [17:18] dbarth, content hub and download API are the critical ones. Non-critical: some way to get the U1 credentials from the system; some way to securely store a token. [17:19] aquarius, we have an OA binding [17:19] aquarius: but i will challenge you to have a layout that is the same as it would on qml [17:19] already [17:19] aquarius: ;) so what would that take? [17:19] dbarth, I can't *have* grid units. I do layout in CSS, because That's What You Are Supposed To Do, and CSS can't do grid units. [17:19] alex-abreu, in an html5 is there a way to control the keyboard? i'm playing with an html5 app that as a text input, but the keyboard covers it [17:19] alex-abreu, yeah, but U1 credentials aren't in OA, I don't believe. [17:19] in qml we can control where it binds [17:19] aquarius: what's missing in the OA api for that? [17:19] dbarth, what's missing is that the U1 credentialsa ren't stored in OA [17:20] aquarius, they aren't ? [17:20] dbarth, but I don't know where SSO credentials *are* stored [17:20] aquarius, yes they are [17:20] i thought they were [17:20] pmcgowan, rly? [17:20] you shouldn't know really [17:20] yes! [17:20] pmcgowan, I can ask OA for my U1 username and password and get it? [17:20] or do I misunderstand [17:20] aquarius, we needed that for the click store [17:20] aquarius: which part of the creds do you need? [17:20] * aquarius looks astounded [17:21] look at your u1 account in system-settings, it should display your username [17:21] aquarius: the identity token, should be there [17:21] kenvandine you may be able to listen for the focus event on the input then preventDefault() on the event object [17:21] there they are, in System Settings > accounts [17:21] ah [17:21] aquarius: the app tokens, should not, should be in your app [17:21] now I need to see whether I can use the app token! [17:21] rock and roll. [17:21] aquarius: and here we have the example of the reminders.app which is using the new api key approach [17:22] aquarius: where apps provide their own app keys to OA for the initial auth. phase [17:23] OK, am now looking into the OA bindings :) [17:23] I wish the HTML console was reflected into the QML console so I could see console.log debugging from inside the SDK rather than having to spin up an inspector! [17:23] kenvandine, not atm (keyboard), but I could add a binding ... just ask & file a bug :) [17:23] hatch, i want the keyboard to show, but i want to ensure the UI isn't covered [17:24] will do [17:24] aquarius, thats the idea behind the qtc integration that we plan to work on [17:24] i want it to do what the qml does in our sdk [17:24] kenvandine ohh yeah that's a bug, it should scroll into view [17:24] aquarius: that inspector window is next on my list [17:24] wicked [17:24] aquarius: weirdly i couldn't get it to work last time i hacked on qtc-plugin [17:24] but qtwebkit should be able to run that i'm told [17:25] hatch, it could just be something about the app, i just modified an existing html5 app [17:25] bah! [17:25] aquarius: we'll talk about that with kyleN in the next session i guess [17:25] what event do I have to wait for, for window.external to become available? [17:25] not sure if i can change something in the html to make the platform dtrt [17:25] kenvandine well typically (on other platforms) when the keyboard opens, the focused element is scrolled into view above the keyboard [17:26] if there is a 'keyboard-open' event you could animate the scroll I suppose :) [17:26] erm. [17:26] I can say "run html5 application on device" [17:26] how do I *stop* it? [17:26] aquarius, upstart-app-stop ? [17:26] :-D [17:26] close application on device is greyed out [17:26] aquarius, 'ubuntu-webapps-api-ready' is the evnt name [17:27] aquarius, a running app on device displays on the Applications page [17:27] hatch, so maybe in this app the element i care about isn't really focused? [17:27] alex-abreu, fired on window? document? [17:27] aquarius, document [17:27] long press on it causes an 'X' in top left corner [17:27] aquarius, this is abug then (if stops works for qml apps) [17:27] kyleN, yeah, but I should be able to kill it from Ubuntu SDK :( [17:27] yes :) [17:27] alex-abreu, yeah, stop works for qml apps [17:27] kenvandine I'm not sure, it's entirely possible it is a bug [17:28] if I run-on-device from Ubuntu SDK, then it lets me run-on-device again, which suggests that Ubuntu SDK isn't registering corectly that the app *is* running [17:28] aquarius, its a bug then, you can file one in qtcreator-plugin-ubuntu LP [17:28] aquarius: file a bug, i monitor those daily [17:29] uh, what alex-abreu said [17:32] will file a bug shortly [17:34] holly sheet. [17:34] there's my U1 account. [17:34] Nice. [17:34] How do I get the token from it? [17:34] http://developer.ubuntu.com/api/html5/sdk-14.04/OnlineAccounts.Account/ is... not the most detailed piece of documentation I have ever seen ;) [17:37] https://bugs.launchpad.net/qtcreator-plugin-ubuntu/+bug/1290965 filed, dbarth [17:37] Launchpad bug 1290965 in qtcreator-plugin-ubuntu ""Run on device" for HTML5 apps not working right; app cannot be stopped" [Undecided,New] [17:42] YEAAAH I have u1 credentials. Rock and roll. [17:42] man, that's excellent. [17:43] OK, I can write a U1 app. Don't need to do sign in. SUperb. [17:48] aquarius, yeah ! :) [17:50] aquarius: https://plus.google.com/hangouts/_/hoaevent/AP36tYdvimJo9VVSLE5Moms6GheyPZT-mtRfIhxxVZq1DOHz-1bXKQ?authuser=1&hl=en [17:52] ? [17:52] omg is it my session? [17:52] ok :) [17:55] aquarius: yes indeed [17:56] if anybody wants to be on the hangout: https://plus.google.com/hangouts/_/hoaevent/AP36tYdvimJo9VVSLE5Moms6GheyPZT-mtRfIhxxVZq1DOHz-1bXKQ?authuser=1&hl=en [17:56] we'll be starting in a few minutes [18:01] aquarius: ok [18:03] I can't join the video session, but I've very interested in an idea like this. I wrote very basic component manager last year called Code Units and I use it all the time on my projects to pull in other code I want to use [18:03] i think choosing an online-account to log into is a good example too [18:03] what about licensing? [18:04] use the pad guys [18:04] http://pad.ubuntu.com/uds-1403-ubuntu-component-store [18:04] iBelieve: pls provide link to code units [18:05] iBelieve: also append it to the pad [18:05] nik90: https://github.com/iBeliever/devutils, it's the code command [18:07] mhall119: ^^ [18:14] I agree with mhall [18:14] aquarius: but what if i *want* to write malicious apps using malicious components?! [18:15] mhall119, aquarius: the "gatekeeper" doesn't have to be human [18:15] dobey: what could it be? [18:15] a bunch of scripts [18:15] it's not like humans can read compiled binaries anyway [18:16] well, the click store has scripts that does certain checks [18:17] Can we not test the components before putting in the component store? [18:17] automated? [18:18] what about a rating system similar to apps, where if someone finds a problem with a component they can give it 1-star and mention that it's spyware or whatever [18:20] aquarius: you don't need a web UI for ratings/reviews [18:22] ucs rate 1 'this thing is rubbish' [18:23] it wouldn't be difficult to build in python [18:23] that's rubbish [18:23] it would bundle it in your source right? [18:24] it needs a central store, and has to be delivered over a secure protocol like HTTPS [18:24] especially if it's going to accept binaries [18:25] they shouldn't pull directly from soruce branches... make them download an archive with a particular structure [18:25] s/soruce/source/ [18:25] yeah [18:25] so released archives [18:25] would be preferable [18:25] you don't want people shoving binaries into VCS all the time [18:25] yeah... and i want to be able to specify the version of the component [18:25] not tip [18:25] aquarius, ^^ [18:26] aquarius, mhall119: yes, you can grab a tarball off launchpad [18:26] without manually having to upload one [18:26] dobey, sweet [18:26] but i still want reliably versions... real declared stable releases [18:26] so if i pull dobey's component and know version 1.0 works [18:27] i don't want to get a newer version of 1.0 [18:27] aquarius: bzr export lp:foo blah.tar.gz [18:27] LP let's you download a tarball of a bzr rev [18:27] yes, you can download a tarball of any rev [18:27] aquarius, but i want to know the downloaded file will be the same if i download 1.0 next week [18:28] So is there any type of quality control from Canonical or just from the community? [18:28] dobey, are you volunteering to implement that in LP ? [18:28] aquarius, https://bazaar.launchpad.net/~unity-team/unity-scope-mediascanner/trunk/revision/67?start_revid=67 look for "download tarball" [18:28] true [18:28] LP tarball links look like: https://bazaar.launchpad.net/~andrewsomething/typecatcher/trunk/tarball/178 [18:29] so we trust the maintainer [18:29] yeah, taht [18:29] basically i'm saying i want you to provide a link to a versioned archive [18:29] and that's what the store has [18:29] and each new version has a different url [18:29] but no smarts built into the store, just a link :) [18:29] in the manifest or something [18:30] aquarius, i think we're saying the same thing [18:30] so who cares if it comes from github or LP [18:30] i do. [18:30] dobey, you're special :) [18:30] you can even link just to the tip of trunk https://bazaar.launchpad.net/~andrewsomething/typecatcher/trunk/tarball [18:31] I think this is a extremely good idea. Think how easy it will be to get new developers or wannabe developers. [18:31] kenvandine: i just don't want it to be arbitrary URLs, because it means people will use insecure URLs [18:31] so you want to enforce https [18:31] oh, we should include an sha1sum or something in the manifest [18:31] valid https, yes [18:32] because the original component might not be malicious, but if someone MITMs your download and sends you malicious code instead, well :) [18:33] i really don't care if archive downloads happen over https... there isn't sensitive data, but we need to be able to verify the downloaded archive [18:33] dobey, then verification fails [18:33] kenvandine: how would it verify it? [18:33] sha1sum [18:33] oh... i guess that would be up to the developer [18:33] kenvandine: but where is that delivered from? [18:33] unless we had a wrapper tool that did the download [18:34] the manifest when the component is submitted to the store would include a url to the archive and the sha1sum [18:34] not a url to a sha1sum [18:34] so we can reliably verify the downloaded archive matches what was submitted to the store [18:35] and if the archive changes at the same url, it fails [18:35] well, i'd do sha256 or something (since sha1 is broken), but as long as it's coming from a secure stream, it's acceptable [18:35] sure... anything like that [18:38] aquarius: ucs could provide a CMake module and just be an integrated part of the build process [18:39] if you don't verify whether the component changes, then you should include them in the source [18:39] aquarius: so just have the components.json and ucs pulls them when you build [18:39] woah! http://bazaar.launchpad.net/~sil/+junk/ucs-demo-app/tarball/2 works. [18:40] Thank you dobey, ssweeny, asomething for correcting me :) [18:40] told you :) [18:41] :) [18:45] mhall119: solr would probably be fitting [18:46] aquarius: is that black brick next to you the new art piece? [18:47] to your left in front [18:47] there's a cube [18:48] aquarius is too excited at the moment! [18:48] yes [18:48] you asked him about his computer! [18:48] lol [18:48] it looks pretty black [18:52] aquarius: you can write it in c++! [18:56] time is up [19:00] balloons: ping? can you ping us the hangout link when ready? [19:00] dbarth, https://plus.google.com/hangouts/_/hoaevent/AP36tYec1w53epatZ3C8nZ_FapqL0EQVY0KX0Ef223gpehp6sIemEQ?authuser=0&hl=en [19:01] dobey: want to jointhe API website discussion? [19:02] https://plus.google.com/hangouts/_/hoaevent/AP36tYd0F4PHPAhxJ3WVvkHEiV5W9xHctARZHR937CY7NulnlfFDcg [19:02] balloons: is this the core apps review session? [19:02] for anyone wanting to join the core apps review session ^ [19:02] nik90, [19:02] yes [19:02] mhall119: i wasn't planning to, but i can [19:04] core apps devs you guys here? [19:04] m-b-o1: ^^ [19:04] http://pad.ubuntu.com/uds-1403-core-apps-review [19:06] dpm, nik90, sorry, dinner in ~15 minutes here :/ [19:07] rpadovani, no worries :) [19:38] iBelieve: anythings to mention for the file manager [19:38] iBelieve: that's next on list [19:39] nik90: thanks, I'll add a couple important items for File Manager that we're currently working on [19:40] nik90: or is just dam adding stuff to the pad? [19:40] nik90: **dpm [19:40] iBelieve: you can go ahead [19:40] nik90: ok [19:41] iBelieve, that'd be excellent, thanks! [19:43] dpm: done [19:44] iBelieve, awesome, thanks! [19:44] iBelieve: Are you guys planning to add keyboard shortcuts? [19:44] nik90: some work, (such as Ctrl+L), but yeah, that is important [19:46] dpm: also just added a link to our team's Trello board [19:56] iBelieve, perfect, I mentioned it on the hangout too. Looking great! [19:57] dpm: thanks for mentioning it [19:59] np :) [20:00] thanks everyone, a very successful start to UDS! \o/ === ChanServ changed the topic of #ubuntu-uds-appdev-1 to: Currently no events are active in this room - http://summit.ubuntu.com/uds-1403/appdev-1/ - http://irclogs.ubuntu.com/2014/03/11/%23ubuntu-uds-appdev-1.html